Automating low volume/high mix lathe work.

[Jashley73]

If we're talking about automating of "chucking" type work, then nothing can compete with an inverted-spindle, vertical lathe. (Yes, spindle facing down.) When you watch a 2-axis lathe load & unload itself, using only the X/Z axes & chuck opening/closing, you have a Holy crap..." moment for sure.

I have not seen them with a sub-spindle, but I have seen them with tailstocks & steady rests.




Tony is correct though. Weather you guys use a robot or not, you need to get into parametric/macro programming ASAP, especially if you are running batches of similar-part families.




CNC-Drehmaschine VL 6 fur die automatisierte Bearbeitung - EMAG Videos - YouTube

 

[adamm]

If we're talking about automating of "chucking" type work, then nothing can compete with an inverted-spindle, vertical lathe. (Yes, spindle facing down.) When you watch a 2-axis lathe load & unload itself, using only the X/Z axes & chuck opening/closing, you have a Holy crap..." moment for sure.

I've never seen one of those before. That is awesome. It's like the idea of using a VMC as a lathe, except it is actually designed for the work.

 

[Gobo]

If a UR10 will handle the parts for load and unload, then they are almost certainly less than three inches in diameter and less than a foot long.

If they are less than three inches in diameter and less than a foot long, a standard bar feeder and parts catcher is much, much simpler "automation" than a robot.

What am I missing?

It will be required to process both slug and bar work.

 

[XD341]

Take a look at Halter CNC automation. It's not a new concept it's just packaged well. It does tend to simplify the material handling aspects of robot machine tending. It is made primarily for lathe chucker work and handles the safety aspects well I think.

I don't own one but I researched them when I bought my most recent lathe.

 

[Comatose]

Right, but your additional process window for slug work is tiny with a 10kg robot. What's the biggest part you'll have to do on it?

 

[Gobo]

Right, but your additional process window for slug work is tiny with a 10kg robot. What's the biggest part you'll have to do on it?

4 inch O.D., app. 6 kilos.

 

[Tonytn36]

4 inch O.D., app. 6 kilos.

You are already overweight / undersized if this is the case. Remember... you need to pick up a finished part AND have a raw part in the gripper assembly. Else you will be slow as molasses.

Robot payload means anything after JT6. So this means the adapter plate, any other plates, bolts, air fittings, washers , nuts, dowel pins, GRIPPERS, Gripper Arms, Gripper Pads, sensors, that are on your EOAT from the flange of JT6 out.

 

[Gobo]

You are already overweight / undersized if this is the case. Remember... you need to pick up a finished part AND have a raw part in the gripper assembly. Else you will be slow as molasses.

Robot payload means anything after JT6. So this means the adapter plate, any other plates, bolts, air fittings, washers , nuts, dowel pins, GRIPPERS, Gripper Arms, Gripper Pads, sensors, that are on your EOAT from the flange of JT6 out.

This process is driven less by cycle time than the desire to reduce labor cost. We are willing to accept a longer cycle time if we are able to meet our production goals. Operator free production where feasible and sensible is one target.

 

[Jashley73]

It will be required to process both slug and bar work.

You have 17 other lathes - some with bar feeders - and you can't dedicate bar-fed work to one of those 17?


Regarding automating of turning work, you really need to take a hard look at priorities. It's difficult to accomondate both chucking work, AND bar-fed work completely unattended, with short cycle times, without spending a ton of time & effort on automation. If you can isolate one machine to chucking work only, or bar-fed work only, things get much, much simpler.

If you try to take on the project of "one lathe, all mix of parts, everything automated" then you have a very steep hill to climb.

 

[Gobo]

You have 17 other lathes - some with bar feeders - and you can't dedicate bar-fed work to one of those 17?


Regarding automating of turning work, you really need to take a hard look at priorities. It's difficult to accomondate both chucking work, AND bar-fed work completely unattended, with short cycle times, without spending a ton of time & effort on automation. If you can isolate one machine to chucking work only, or bar-fed work only, things get much, much simpler.

If you try to take on the project of "one lathe, all mix of parts, everything automated" then you have a very steep hill to climb.

Some of our equipment is from the early 80's and will be retired and replaced with newer equipment. Your statement about the steep hill to climb is an argument I have made repeatedly. I wholeheartedly concur. When we get machine tool salespeople in, their thoughts are the same. But how do you get the decision makers to understand? I am told what is wanted, and have to produce - something. Tilting at windmills is what it sometimes feels like. They want a magic machine that will do it all. I know it wont happen, but have to give them something.
Setting all that aside, I am just looking for any good ideas from you smarter than me people. Several good ideas have already come up, and I appreciate that.

 

[Comatose]

You can't replace 17 machines with one, though.

A robot plus integration to get it running well is about a $100k project.

$100k is the difference between a lathe with a 3" spindle bore and a 4" spindle bore.

What you really want is three lathes: one with a 2.6 inch spindle bore and a bar feeder, one with a 4" spindle bore and a bar feeder, and one with a robot set up as a chucker.

If they want to automate something, doing software automation for job routing to efficiently use those three lathes for the job at hand will put you MILES ahead of having three eierlegendewollmilchsaus being utilized inefficiently.

Now, your bosses want to see a robot fiddling around with parts and I get that. Robot cells dog-and-pony show beautifully. So make the chucker one first, then start stomping for the other two that you need.

All we're saying is your robot isn't big enough. You need a load gripper and an unload gripper. You need an adapter plate to mount the load and the unload gripper. There's no way with an anything-that-comes-down-the-pike mandate you'll be able to always use one gripper for both. You need sealed grippers.

One thing that I've recently discovered that may save you a ton of time and cost is that the continuous carbon filament reinforced 3d printed output from one of the Markforged machines is entirely stiff and durable enough for medium production gripper fingers, and absurdly cheaper than making metal ones. Lighter, too.

 

[Gobo]

You can't replace 17 machines with one, though.

A robot plus integration to get it running well is about a $100k project.

$100k is the difference between a lathe with a 3" spindle bore and a 4" spindle bore.

What you really want is three lathes: one with a 2.6 inch spindle bore and a bar feeder, one with a 4" spindle bore and a bar feeder, and one with a robot set up as a chucker.

If they want to automate something, doing software automation for job routing to efficiently use those three lathes for the job at hand will put you MILES ahead of having three eierlegendewollmilchsaus being utilized inefficiently.

Now, your bosses want to see a robot fiddling around with parts and I get that. Robot cells dog-and-pony show beautifully. So make the chucker one first, then start stomping for the other two that you need.

All we're saying is your robot isn't big enough. You need a load gripper and an unload gripper. You need an adapter plate to mount the load and the unload gripper. There's no way with an anything-that-comes-down-the-pike mandate you'll be able to always use one gripper for both. You need sealed grippers.

One thing that I've recently discovered that may save you a ton of time and cost is that the continuous carbon filament reinforced 3d printed output from one of the Markforged machines is entirely stiff and durable enough for medium production gripper fingers, and absurdly cheaper than making metal ones. Lighter, too.

Well said. Great advice on the printed grippers.

 

[garyhlucas]

Some of our equipment is from the early 80's and will be retired and replaced with newer equipment. Your statement about the steep hill to climb is an argument I have made repeatedly. I wholeheartedly concur. When we get machine tool salespeople in, their thoughts are the same. But how do you get the decision makers to understand? I am told what is wanted, and have to produce - something. Tilting at windmills is what it sometimes feels like. They want a magic machine that will do it all. I know it wont happen, but have to give them something.
Setting all that aside, I am just looking for any good ideas from you smarter than me people. Several good ideas have already come up, and I appreciate that.

You face the same problem I have. You are an employee trying to tell the boss that he is wrong. You are right but as an employee you can never know more than the boss. So you need a really expensive consultant they can’t afford to ignore without looking foolish. Have that consultant present the ideas you know are workable, like bar feed work on lathes that do only barfeed, chuckers doing only chucker work. When you implement his suggestions and it works you both look like heros. This is what a good consultant does. Indentify the employees that already know how to solve the problem and enable them.

 

[kustomizingkid]

For the bar fed lathes are you running the same materials or have material dedicated lathes? Just thinking what is the advantage of have dedicated material lathes and loading up the bar feeders with material that covers a range of parts where maybe you are running less than ideal material sizes and make pointless chips with the benefit of just filling the bar feeders and making parts... cycle times are less important right?

 

[Gobo]

For the bar fed lathes are you running the same materials or have material dedicated lathes? Just thinking what is the advantage of have dedicated material lathes and loading up the bar feeders with material that covers a range of parts where maybe you are running less than ideal material sizes and make pointless chips with the benefit of just filling the bar feeders and making parts... cycle times are less important right?

Please see my post #16. To flesh this out a bit, we might make 3 different pistons with very similar features and geometry out of the same material on 3 different machines. Why -you may ask. Because they go into 3 different product lines and someone thinks that is a good idea. Not my decision, I just have to work within that strategy,face the challenge presented and do my best.

 

[Jashley73]

But how do you get the decision makers to understand? I am told what is wanted, and have to produce - something.

..........

They want a magic machine that will do it all.

This is going to be a long post, but bear with me...




Everyone wants the one machine to do it all.


How do you convince them? You have to show them.

Get a quote for...

Japanese 12" chuck CNC Lathe with 3" bar capacity, sub-spindle, y-axis, live tools, parts-catcher, bar feeder, 65kg robot with multiple end-of-arm tools to cover the entire range, safety perimeter guarding, turn-key integration for the whole shebang, and a couple months of paying the machine dealer's AE to program EVERYTHING like they want.

$450-500k easily. Go mill-turn, and you're pushing $700-800k...

Build a tool library to run EVERYTHING they want through that machine. Now, you have to fit it on a 12-station turret. Figure $4k for a 90* head live-tool, $2k for a straight live-tool. Capto tools are the only way to go for quick-change-over. The receivers are about $1k a piece, the heads are $250-300 a piece.

Say it takes about 3 minutes per turning tool to swap the Capto heads, and touch-off to the tool-setter. About 8 minutes a piece on the live tools. 5 minutes to change chuck-jaws. 10 minutes to change the robot's end-of-arm tooling. 10 minutes to change programs & verify offsets. 10 minutes to dial in the 1st part.

Then, do a time study on the cycle times of some of your high-runners. Make sure you do a bar-fed part, and a chucker part.

What you need to do then, is take... Total production time (5 days a week x 1/2/3 shifts) - Setup time = available production time x .75% efficiency = your total production capacity time. Remember to add extra setup time, to accommodate changing over between bar-fed jobs, and chucker jobs. Now, divide that production capacity time by part cycle time. Remember that cycle times will be a mix of bar-fed & chucker jobs.

What you'll end up with, is XXXXX chucker parts, and YYYYY bar fed parts produced per day/week/month/year, at $$$$$$$ capital investment.


There's the cost of their fantasy machine.






Now, to compare....


Get a quote on a machine for chucking work - a 12" CNC with y-axis live tools, and a sub-spindle if your work requires, plus the 65kg robot. My estimate is $350k for a Okuma/Mazak/Mori + Fanuc Robot, turn-keyed. Run the same exercise, but this time ONLY with the chucker-parts. Now, you have XXXXX chucker parts produced with $$$ capital investment.

You might even consider one of the inverted spindle lathes for the chucking work too, and eliminate the robot altogether...

And another quote on a 8-10" CNC with a collet nose, barfeeder, parts-catcher & live tools, y-axis & sub if you need them. $250k? Or even a big swiss machine if the parts will fit. Run the same exercise but only for the bar-fed parts. Now, you have YYYYY bar-fed parts for $$$ capital investment.




It is a lot of work, but it will be worth it. The winner might be surprising, or it may be tighter than you think.

I have some tools to help with all of this, but I'd probably have to charge in the "high priced consultant that's too expensive for them to ignore without looking foolish" range to take on a project such as this, in Oregon...



Best of luck - Don't be scared - you just have to try to *think* like them, and then be prepared to show them some data that they can't ignore.

 

[Tonytn36]

My advice is only offered because I have a bit of experience in this arena. I've designed, built, implemented, programmed and supported a bit over 200 robotic cells since 2004, the vast majority servicing machine tools. By designed, I mean I designed the entire loading process and in most cases the fixturing of the machine and any in-machine automation to assist with the robot loading. Designed the end effector of the robot which in some cases is quite complex (changing spacings of multiple grippers, handling packaging materials, vision systems, multiple tools (EOAT changing), all types of sensors, etc). Designed the placement and mounting of the robot and in some cases, the structure to be added to the machine tool to mount the robot. Designed the safety enclosure, and external automation (conveyor systems, drawer systems, dunnage feeding systems, orientation fixtures or flippers, etc). So I do write with a bit of understanding of these systems and what is required to implement one successfully.

If you do not do this correctly up front, it will be a nightmare to deal with later.

The point I make about the end effector weight is a very valid one. Any gripper that will handle a 6kg part is not going to be light. For instance, a Schunk DPG-160-1 - which is what I would recommend for such an application - will handle the workpiece weight with a max of 7.8 kg and a 16 mm per jaw stroke, but the gripper itself (no gripping arms or pads) weights 3.65 kg. So you are probably going to be over 5kg with just the gripper, arms and pads and fasteners to bolt the arms on with - note that you still have to attach it to the robot and run air lines, fittings and sensors / cables to it. You will likely be over 6kg in EOAT with no part and thus overweight with your max 6kg part on a 10kg robot.

Either the robot needs to get bigger or the requirements must change to limit the weight of the parts which in turn probably lightens the EOAT because you can use a smaller gripper or a lighter gripper with less gripping force such as a PSH style.

For any project I do, I start an Excel spreadsheet for EOAT weight. Every piece is weighed, either a calculated weight from the 3D model, or physically weighed. In addition, you must adhere to limiting to the max torque for JT5 of the robot. Thus you must watch your EOAT lengths and where the mass is within the motion path of the robot.

There is more to it than most people think.

 

[CarbideBob]

.....

There is more to it than most people think.

Oh it's easy, just pick and place.
A min wage human can do it, how hard can it be.... all those calculations, spreadsheets and weird numbers are just to keep engineers employed.
....

The thing with robots or any automation is not making it work for the first days or weeks which is easy and you can do much wrong and burn many rules during this phase, it's making it work for years without failure which is why you bought in.
Bob

 

[Miguels244]

Bar...already done in process control.
Hard part is changing stock, you can segregate stock to ease that problem.
Operator assist to change the bar feeder.
Make sure you have an unload chute.
It will take a person full time to manage the moving parts...and a good one.
The money saved will be in efficiency, but the guy who can do that job ain’t cheap.

 

[Mechanola]

Bob, my feelings are with you. As others and you yourself have said management is the problem. As long as the designer does not adapt to production, production is adapted to design. This sounds foolishly poetic but it is the hard truth. They can buy machine over machine and never grasp that their thinking goes one way only. Totally un(der)developed brainpower. They’re looking for a machine on which one can mill internal gears of two inches diameter. Poeticly foolish